I am trying to derive Land Surface Temperature(LST) from Sentinel 3 SLSTR dataset. The given literature talks about split-window and dual angle algorithms - Link

Ts = Ti + c1(Ti- Tj) + c2 (Ti- Tj)2 + c0 + (c3 + c4W)(1 - εn)+(c5 + c6W)Δε

T = at-sensor brightness temperature at bands “i” and “j”
W = total atmospheric water vapor 
ε = mean emissivity 
Δε = emissivity difference at the two bands
c0-c6 = Derived from statistical simulation

Emissivity calculation can be done as mentioned in the paper itself. However, I am not sure how to go about c0-c6 for split window. They might have carried out some statistical regression on ground/airborne datasets.

Is it possible to simulate MODIS LST and TIR Bands/any other satellite dataset (with dual angle algorithm) and then calculate the coefficients? I may also have access to weather station data for my region.


Basically, your problem is that the parameters c0 to c6are undefined. The authors of the paper you're pointing to have used a physical model to simulate a large set of synthetic observations, and have derived these coefficients by fitting the formula you include to the simulations.

I think they do report these coefficients in Table 1 (confusingly calling them a0 to a6). You can check whether this is correct by e.g. taking a contemporary MODIS LST retrieval and checking that the values are broadly in line with the ones you retrieve from using the split window algorithm.

  • But then these coefficients would be applicable to the mapping done for a different region of interest than mine. – fjord May 6 '17 at 6:42
  • Moreover, I don't have access to MODTRAN Software hence I was wondering if a Look Up Table could be generated using python/R in order to map the atmospheric radiosonde data for my region of interest. – fjord May 6 '17 at 6:44
  • @fjord No, the coefficients are supposed to cover all types of atmosphere types and emissivities. You don't need MODTRAN, just the numbers on the table. As you can see from the paper, the algorithm performs on vastly different land cover types/regions, so you should be able to just use it. – Jose May 7 '17 at 12:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.